Control over the Theory: Reforming the icc’s Approach to Establishing Commission Liability?

The icc has employed the ‘control-over-the-crime’ theory, which treats those who ‘control’ the commission of a crime as principal perpetrators. Legal academics and icc judges have criticised the Court’s reliance on that theory for producing unsound legal reasoning. This article engages with the question from a novel perspective, that focuses on the institutional factors affecting the adoption and reform of legal theory. Transplanting Barnett and Finnemore’s concept of the ‘pathologies’ of international organisations to the field of international law, the article argues that reforming the rules for assessing criminal responsibility is a challenging endeavour, even when those rules have exhibited significant deficiencies. Reform is possible, but it is more likely to be incremental rather than revolutionary. The findings also bear implications for international criminal justice more generally, as they suggest that the answer to delivering sound judgments is not improving criminal law theory but appreciating the peculiarities of each case.


Introduction
International criminal law (icl) has historically faced the dilemma of determining what it means to 'commit' an international crime. Since 2007, the International Criminal Law Review (2022) 1-29 2 International Criminal Court (icc) has opted for the 'control-over-the-crime' theory which, as its name suggests, requires evidence that the accused had controlled the commission of the crime, regardless of whether they had physically perpetrated it or not.1 This has become the Court's preferred approach for assessing criminal responsibility for committing a crime, as defined in Article 25(3)(a) of the Rome Statute. The 'control-over-the-crime' theory has been evoked in landmark icc cases such as Lubanga, Kenyatta and Muthaura, Gbagbo and Blé Goudé, Ntaganda, and Ongwen. But the 2021 decision by the Ntaganda Appeals Chamber (ac) raised important questions about the costs of relying on this theory. Two influential icc judges sitting at the Ntaganda ac-Judge Chile Eboe-Osuji and Judge Howard Morrison-expressed concern that the 'control-over-the-crime' theory was 'at best unnecessary' and at worst-led to unsound legal reasoning behind icc judgments.2 They are not the first ones to argue against the 'control-over-thecrime' theory-in 2012 similar remarks were made by Judge Fulford3 and Judge Van den Wyngaert.4 Criticism of the Court's use of that theory has also been expressed in many academic publications by influential scholars.5 Specifically, critics have identified several problems resulting from the difficulties of reconciling the abstract requirements of the 'control-over-the-crime' theory with the complex reality of mass atrocities. The practice of making the facts and evidence presented at the Court fit the requirements of that theory has often 3 led to awkward legal reasoning in cases, like Ntaganda.6 Furthermore, the icc has ended up producing the same story (that the accused had 'controlled' the crimes through a well-organised criminal apparatus) in very different cases, such as Gbagbo and Blé Goudé, Ongwen, and Kenyatta and Muthaura.
The problems associated with the 'control-over-the-crime' theory are of crucial significance considering the special status of the icc-it is the only permanent international criminal court and the trend-setter in the field. The criticisms of the theories of criminal responsibility used at the icc and the possibilities for reform should be taken seriously because of the influence of icc jurisprudence in icl and its significance for the future of international criminal justice.
This article engages with the scholarly analysis of the icc's 'control-overthe-crime' theory7 from a novel perspective, that focuses on the institutional factors affecting the adoption and reform of legal theory. It relies on insights from sociological institutionalism8 and interdisciplinary studies of the international legal field9 to explain why international legal bodies routinise inefficient rules and, specifically, to assess the possibilities of reform concerning the 'control-over-the-crime' rule at the icc. Transplanting Barnett and Finnemore's concept of the 'pathologies' of international organisations to the field of international law, this article observes that a ground-breaking reform of the rules for assessing criminal responsibility is a particularly challenging endeavour, 4 even when those rules have exhibited significant deficiencies. The reason is the crucial importance of the predictability of legal rules in international criminal justice. The analysis is supplemented with observations from the operation of the icty, which elucidates how pathologies work specifically with respect to the rules concerning the modes of liability at international courts and tribunals. The failed attempts to dismiss the concept of jce and to reform command responsibility illustrate the difficulties of reforming routinised legal rules.
Reform is still possible, but it is likely to be incremental rather than revolutionary. This article observes some indications, that the icc is slowly moving away from the 'control-over-the-crime' theory. Specifically, the 2019 and 2021 decisions on the confirmation of charges in the Yekatom and Ngaïssona10 and Abd-Al-Rahman11 cases, respectively, both focused on the direct link between the defendant and the crimes, without relying on the 'control-over-the-crime' theory. In 2021 the icc's Appeals Chamber found that approach compliant with the requirement of providing the accused with fair notice of the charges.12 But the analysis shows that dismissing an inefficient rule is only part of the solution. While some judges have suggested relying on other modes of liability, such as Article 25(3)(d), in the future, there is still no clear consensus on the way forward.
Crucially, the analysis suggests, that the problems with the 'control-overthe-crime' theory evidenced in icc jurisprudence should not be understood merely as the shortcomings of a particular theory, but also as a cautionary tale about the danger of over relying on criminal law theory in general when it comes to international crimes. The excessive focus on improving criminal law theory could obfuscate the main challenge to delivering sound judgments, which is the difficulty of collecting evidence of the accused's involvement in the crimes when those persons have been removed from the crime scene. Consequently, the lesson from this story should not be to search for a new theory in attempt to deliver more convincing judgments, but to appreciate the peculiarities of each case and to focus on the quality of the available evidence. 10  Finally, this article contributes to the critical scholarship on international criminal justice which explores the narratives created by icl proceedings.13 Several studies have suggested that icl proceedings create a particular image of the perpetrator of mass atrocities as male, barbaric, often a national of a country in the Global South.14 The review of icc jurisprudence suggests that the reliance on the 'control-over-the-crime' theory has further created a particular image of the dynamics and modes of organisation of international crimes. The Court has ended up portraying very different African conflicts in the same terms: as the product of a few powerful masterminds using nameless troops, including children, as tools to commit atrocities.
This article is structured as follows: Section 2 presents the framework of analysis; Section 3 briefly discusses the routinisation of the 'control-over-thecrime' theory at the icc and the deficiencies of relying on that theory revealed in icc practice; Section 4 discusses the potential for reform of the Court's interpretation of Article 25(3); Section 5 concludes.

Rule-following, Pathologies, and Opportunities for Improvement
This section discusses the institutional factors that affect the development of norms guiding the operation of international organisations, and specifically-of international legal bodies. It borrows insights from the school of sociological institutionalism15 and interdisciplinary studies of the international juridical field.16 The analysis highlights the challenges that the routinisation of rule-following could pose for the efficiency of international courts, but observes, that the dynamic nature of the international legal field constantly renders opportunities for change and improvement. As every other type of international organisation, international courts and tribunals tend to foster specific rules of conduct that guide their everyday operation. Over time those rules become institutionalised as part of the culture of 6 a particular organisation-they obtain a 'taken-for-granted' status, that results in habitual compliance.17 Studies of international bureaucracies have shown, that at least for the duration of their mandates the persons working at an international organisation socialise within its culture and identify with its rules. 18 The development of an internal culture within international courts ensures clarity of the rules guiding its operation. Specifically in the field of icl, this is of crucial importance because it ensures fair notice to the defendants about the applicable rules. While the Rome Statute and icc Rules of Procedure and Evidence mark a significant improvement in that regard compared to the international criminal tribunals for the former Yugoslavia (icty) and for Rwanda (ictr), the codified text always requires interpretation to resolve ambiguities and overlaps between the provisions. The routinisation of rules for interpretation could reduce the uncertainty about the applicable rules in those situations. Another virtue of the existence of a strong internal culture of collegiality among judges is its potential to enhance the perception that the court's legal decisions are authoritative and sound. 19 Overall, the internal culture of international organisations, including legal bodies, constitutes their identity as actors in the international arena.
But the routinisation of rule-guided behaviour could also pose problems for international courts. Specifically, it could lead to the development of what Barnett and Finnemore call 'pathologies' , namely, the prioritisation of rule-following instead of delivering the best possible outcome.20 The habitual compliance with rules obscures the means-end rationality of the persons working at international courts and makes it harder to address the inefficiencies in the system. The principle of legality in icl renders the optimisation of the rules of international legal bodies particularly difficult. The stakes of reforming icl rules are higher compared to those in other types of international organisations because such reforms concern the freedom of the individuals on trial. In fact, the expectations of icl's compliance with the principle of legality have increased with the establishment of the permanent icc and the rise of the influential branch of legal scholarship called the 'liberal critique of icl' ,21 7 which has advocated for greater compliance with criminal law principles in international proceedings.
This does not mean that once a rule becomes internalised in everyday icl practice, it can never change. The international legal field is dynamic and provides opportunities for reform, both from within and from outside international courts. Rules can be challenged from within by individual judges. When judges are selected at international courts, they already have significant experience as practitioners, academics, or diplomats.22 The influence of the personal professional experience on the decisions of some international officials may prove stronger than the court's internal culture23 and lead those persons to challenge the rules followed by their colleagues. Internal court disagreements about the meaning of the law are often manifested by issuing 'dissenting' or 'separate' opinions to a particular judgment.24 The rules followed at international courts could also be challenged from outside observers. International organisations rely on external actors, such as states, non-governmental organisations, experts, and activists, for the successful implementation of their mandates. External actors provide international organisations not only with 'material' resources (technologies, facilities, staff), but also with 'symbolic' ones-they can maintain or challenge the legitimacy of the organisation and its decisions.25 Specifically in the case of international courts and tribunals, decisions become subject to social validation by the broader community of lawyers and legal academics outside the institution, which Schachter famously called the 'invisible college' of international lawyers.26 Law journals, in particular, have become an important venue for debating the rules guiding international legal practice.27 Legal scholarship has served as a 'testing ground' for innovative interpretations of icl and judges have occasionally relied on academic research to validate their interpretations of the law, rendering icl scholarship a 'tool' for fighting legal battles in court.28 icc chambers have often referred to the writings of legal scholars  Nevertheless, even though inefficient rules can be challenged by actors within and outside international courts, any reform of a legal rule is likely to be incremental instead of revolutionary. There are two reasons for this: first, because the reform needs to be balanced against the goal of sustaining the predictability of legal rules, and second, because it requires persuading the better part of the global legal epistemic community in the necessity for reform.32 Any changes in routinised behaviour constitute revisions of the pre-existing socially validated rules rather than completely novel phenomena.33 Furthermore, even if there is an agreement that a particular legal rule is inefficient and requires reform, there might not be an agreement within the international legal community on the precise requirements of the reformed rule.
The rest of the article relies on these insights to explain the continuing use of the 'control-over-the-crime' theory at the icc, despite the significant challenges, that the theory has posed in practice. Section 3 briefly discusses the adoption of the 'control-over-the-crime' theory and the concerns that some icc judges and academic commentators have expressed regarding its efficiency. The overview of icc judgments relying on that theory renders support for the arguments of its critics. Section 4 discusses the opportunities for reform.

Routinising Inefficient Rules: the 'Control-Over-the-Crime' Theory
Article 25(3)(a) Rome Statute names three forms of 'commission' liability: 'as an individual' , 'jointly with another' person or 'through another person, regardless of whether that other person is criminally responsible' . Even though this formulation of the term 'commission' is considerably more detailed compared to the one found in the icty and ictr statutes, it still leaves ambiguities for the icc judges to resolve. Those ambiguities include: how to interpret the terms commission 'jointly with another' and 'through another' person and how to resolve the seeming overlap between the latter provision and Article 25(3)(b) Rome Statute, which concerns a person who 'orders' the commission of a crime.
As is well-known, to address those issues, the judges at the 2007 Lubanga ptc adopted a specific rule-the 'control-over-the-crime' theory.34 That theory recognises as principal perpetrators the persons who 'in spite of being removed from the scene of the crime, control or mastermind its commission' , by deciding where and how the crime is committed.35 Ever since, this rule has become routinised in icc practice-not only in judicial decisions,36 but also in the documents containing the charges submitted by the prosecutor, which often frame the facts of the case in line with the requirements of the 'controlover-the-crime' theory.37 The adoption of the new icc rule promised to avoid the problems associated with its predecessor at the icty-the theory of 'joint criminal enterprise'-and to make the assessment of criminal responsibility at the icc more efficient. jce had triggered concerns for being so broad that it faced difficulties explaining 'why each fish caught deserves punishment for intentional wrongdoing' ,38 as well as for obfuscating the distinction between principals and accessories among the participants in the criminal enterprise.39 The icc's new rule aimed to systematise the main components of the dynamics behind mass atrocities by listing specific requirements for establishing the perpetrator's 'control' over the crime: in the case of joint perpetration-evidence of a 'common plan' ,40 and in the case of indirect perpetration-evidence of a hierarchically organised power apparatus with a sufficient number of interchangeable subordinates,41 and of the accused's ability to 'secure compliance' with their orders to commit crimes.42 The aspiration was that the 'control' theory would, thus, provide clear general rules for the attribution of criminal responsibility and the assessment of blameworthiness, that would be applicable to all cases. Furthermore, the 'control' theory aimed to communicate to the victims and the broader international community 'who was the "real" culprit' behind mass atrocities,43 namely, the mastermind who used 'ordinary people' to commit the crimes.44 However, the dubious notion of 'control' led some legal scholars to express concerns that in practice the new icc rule might reproduce the same deficiencies exhibited by jce.45 A look at the 14 years during which the icc judges and prosecutors have relied on the 'control-over-the-crime' theory renders support for this proposition. This section discusses two inefficient outcomes that have resulted from the icc's routinised reliance on the 'control-over-the-crime' theory and triggered criticism and calls for reform from within and outside the Court: making the facts of the case fit the nebulous notion of 'control' and obfuscating the differences between cases.

3.1
The Nebulous Notion of 'Control' After the adoption of the 'control-over-the-crime' theory at the icc some legal scholars expressed doubts that in the context of mass atrocities one could make a convincing assessment of whether a person had 'controlled' the direct perpetrators of the crimes.46 The 'control-over-the-crime' theory appears to 39  suggest a 'yes' or 'no' answer to that question while, in reality, domination is a matter of degree.47 Several icc judges have raised similar concerns. In 2021 in Ntaganda, Judge Morrison and Judge Eboe-Osuji criticised the significant stretching of the term 'control' that was required to make the facts of individual cases fit the legal theory.48 As observed by Judge Eboe-Osuji, it is not easy to control a piece of 'mechanical equipment' in the way required by the 'control' theory, 'let alone an aggregate collection of a human multitude under a headman' .49 Another problem, which was highlighted by Judge Ušacka in 2009 in Al-Bashir, concerned the 'locus' of control over the crime. The judge's comments suggested that in the case of mass atrocities, it often remained unclear whether a single person had exercised control over the crimes, or whether (and how) control had been shared within the criminal apparatus.50 More recently, in 2021 Judge Morrison observed that under the 'control' theory a broad variety of persons, ranging from leaders to mid-level commanders and direct perpetrators, could be 'equally' treated as principal perpetrators of the same degree of blameworthiness.51 The challenges of delineating the structures of control in the case of mass atrocities has led some icc judges to suggest that the 'control-over-the-crime' theory is 'dangerous' for the Court because it leads to unsound reasoning which damages the authority of the judgments.52 icc judgments relying on that theory have sometimes triggered such criticism. A well-known example is the first icc judgment, which convicted the leader of the Union des Patriotes Congolais (upc) Thomas Lubanga for co-perpetrating along with other senior figures at that organisation the crimes of conscripting, enlisting and using child soldiers to participate in hostilities during the 2002-2003 conflict in Ituri, Democratic Republic of the Congo (drc).53 Some legal experts expressed concern that the judges had inferred Lubanga's control over the crimes by view of his official position rather than his actual involvement in the crimes.54 The icc judges' findings concerning the accused's 'control' over the crime seem most sound in the cases of what Judge Morrison calls 'simple criminality'58-situations that involve a limited number of persons committing specific crimes, such as the Al-Mahdi case. The case concerned the intentionally directed attacks against religious and historic buildings in Timbuktu. Crucially, the tc relied on video evidence of the accused's personal participation in the destruction of five buildings of protected cultural heritage59 and Al Mahdi's admission of his own guilt.60 Thus, the direct link between the accused and the crimes was clearly established. Similarly, Judges Morrison and Eboe-Osuji did not question the tc's findings concerning the crimes that Ntaganda had perpetrated himself, but only the findings that crimes committed by Hema civilians were attributable to Ntaganda because of his alleged control over their actions. These cases demonstrate that the findings under Article 25(3)(a) sound significantly more convincing when the proximity between the accused and the 55  13 crime is greater, than similar findings in cases concerning complexly organised power structures where the notion of 'control' becomes nebulous. The problem is that most of the cases that the icc deals with are of the latter kind.

3.2
Obfuscating the Differences Between the Cases Another deficiency of the 'control-over-the-crime' theory is that, as Judge Morrison has noted, it treats 'very different scenarios in the same way' .61 When applied in practice at the icc, the 'control-over-the-crime' theory has ended up portraying a dichotomous picture-of 'masterminds' orchestrating the crimes and 'tools' committing the crimes. This image of the bureaucracy of mass atrocities might have been apt for describing Nazi criminality, but legal scholars have noted that it appears less attuned to modern-day African conflicts that have so far constituted the bulk of icc investigations.62 The icc has tried to avoid that problem and improve the efficiency of the 'control-over-the-crime' theory. The Katanga and Ngudjolo ptc modified the theory, by specifying another mechanism, apart from the replaceability of the subordinates in criminal bureaucracies, that could enable automatic compliance with the mastermind's orders. According to the judges, the existence of 'intensive, strict, and violent training regiments' , including the abductions of minors and their training to shoot, rape and pillage, could similarly ensure the indirect perpetrator's control over the organisation.63 Some legal scholars welcomed the icc's decision to reform what was perceived as the outdated bureaucratic version of the 'control' theory.64 But an overview of icc cases reveals that, contrary to their expectations, the modified version of the theory has, nevertheless, ended up reproducing a simplistic and generalised narrative of violence in Africa. The only difference is that the element of the 'bureaucratic murder apparatus' has been substituted with those of child soldiers and strict and abusive training regiments.
Consider the charges which the icc prosecutor brought in Gbagbo and Blé Goudé, Kenyatta  Crucially, the organisations that had allegedly committed the crimes subject to those cases also significantly differed in terms of their composition and mode of operation. The 'pro-Gbagbo forces' , that supported the former Ivoirian president's claims to power, were allegedly composed by a variety of actors, including military personnel, armed youth, militias and mercenaries. 69 The crimes listed in the charges in Kenyatta and Muthaura were allegedly perpetrated by an organisation called the 'Mungiki' and youth supporting the Kenyan Party of National Unity (pnu).70 By contrast, the Lord's Resistance Army (lra), which Dominic Ongwen was part of, was an insurgent organisation, organised around strong hierarchical relationships and a cult towards the personality of its leader Joseph Kony. 71 The upc was also an insurgent organisation, but it operated in the drc with joint political and military profiles.72 The nuances of each case were largely lost when presenting the narrative through the lenses of the 'control-over-the-crime' theory. In all cases the suspects were said to have participated in the alleged crimes in the same, very specific, manner. All individuals were said to have exercised 'authority' and 'control' over a criminal organisation that had committed the crimes. 73 The organisations were also presented similarly: as 'organised' and 'hierarchical' power apparatuses. 74 It was often specified that the physical perpetrators of the crimes were lower-level members of those organisations, whose personal conducts or wills were of little significance to the overall criminal campaign. 75 The structure and size of those organisations were said to have secured almost 'automatic compliance' with the criminal orders of the person in control. 76 In fact, the documents containing the charges in Gbagbo and Ntaganda presented in surprisingly similar terms the reasons why the forces under their control had guaranteed 'compliance' with the accused's orders. Both documents specified that the physical perpetrators of the crimes had been 'well armed' , including with heavy weapons, had acted in groups composed of 'dozens' of persons, and had received 'military training' . 77 The language is so specific that one may get the impression, that the icc prosecutor is describing the same case, despite the different socio-political dynamics of the violence that have taken place following the 2010 elections in Côte d'Ivoire and the 2002-2003 Ituri conflict in the drc.
This tendency to produce similar narratives has important implications because it creates a very specific perception of the gravest acts of concern to all humanity. Several scholars have expressed concern that international criminal justice has produced a particular image of the perpetrator of mass atrocities as non-white, male, barbaric, often a national of a country in the Global South. 78 The review of icc jurisprudence suggests that the reliance on the 'control-over-the-crime' theory has further created a particular image of the dynamics and modes of organisation of international crimes. The complex socio-political context of mass atrocities has been reduced to a hierarchically 16 structured organisation with top-down command. The functions of different participants in mass atrocities have largely remained obscured and the focus has shifted onto two categories: 'masterminds' behind the crimes and 'tools' for committing the crimes.
One might object, that it is simply not the purpose of a criminal law theory, such as the 'control-over-the-crime' one, to reflect the variety of contexts within international crimes occur. Indeed, icc judgments tend to be very detailed, often containing several hundred pages of information about the context of the criminal events. The problem is that this detailed information is often framed in line with the 'control-over-the-crime' theory. One can surely go into the particularities of a case, but the overall impression is that of the same story: a powerful individual pulling the strings behind the commission of the crimes by dozens of subordinates who are highly obedient, whether out of loyalty or fear for punishment.
One might further correctly point out, that the facts of individual cases always must be made to fit within the framework of the legal process, whether that process relies on the 'control-over-the crime' theory or something else. While this is certainly the case, the 'control' theory has proven particularly problematic in that regard because of the level at which it seeks to categorise the elements of indirect (co-)perpetration. For that reason, as the next section shows, some icc judges have suggested relying on alternative modes of liability that employ more flexible language.
Overall, the icc jurisprudence has revealed, that the excessive categorisation and systematisation of legal rules has not led to a significant improvement in the quality of legal reasoning behind judgments. The complex requirements of the 'control-over-the-crime' theory have led to questionably broad interpretations of the term 'control' in cases where the relationship between the accused and the crimes is distant, and to the assimilation of different cases under a single abstract framework. The next section discusses whether the icc is likely to continue to rely on the 'control-over-the-crime' theory despite these difficulties.

4
The Way Forward 4.1 Challenges Despite the criticism, which the 'control-over-the-crime' theory has endured by both academics and several icc judges, and the shortcomings that theory has displayed in practice, dismissing it altogether is unlikely to be an easy process. Once a norm gets institutionalised in everyday practice, it becomes very hard to change, despite potential inefficiencies in its performance. This section shows that many international judges have prioritised the predictability of legal rules despite observable deficiencies in the use of those specific rules. The discussion focuses on attempts to reform rules concerning modes of liability at the icty and the lessons that could be drawn with respect to potential reform of the 'control-over-the-crime' theory at the icc.
One notable example of the difficulties of reforming modes of liability is the unsuccessful attempt of some icty judges to replace jce with a control-based theory of principal liability. In 2003, four years after the articulation of jce at the Tadić ac, the Stakić tc substituted jce for the German-influenced doctrine of indirect co-perpetration, which posed the following requirements: (a) an explicit agreement or silent consent to accomplish a common goal, (b) coordinated cooperation, and (3) joint control over the crime.79 The Stakić tc considered that their theory of co-perpetration came 'closer' than jce to the plain meaning of the term 'committing' under Article 7(1) and, hence, was more compliant with the legality principle.80 The tc appeared influenced by the German legal tradition, citing the work of the legal scholar Claus Roxin.81 But on appeal, in 2006 the Stakić ac terminated the attempt to substitute jce with indirect co-perpetration. The ac judges argued that the introduction of a new mode of liability should be consistent with the tribunal's jurisprudence, lest it generated 'uncertainty, if not confusion' about the applicable law.82 The ac considered that unlike co-perpetration, jce constituted a mode of liability which was 'routinely applied in the Tribunal's jurisprudence' .83 This reasoning is a clear illustration of the prioritisation of predictable rule-following over optimising procedure at international legal institutions.
Another example of the difficulties of reforming the rules concerning forms of criminal responsibility without having reached a significant consensus is the command responsibility principle. The icty has generally recognised command responsibility as a sui generis form of responsibility where 'the superior cannot be considered as if he had committed the crime himself' ,84 even though the crime of 79  judges, which established 'a new majority of appellate thought' at the tribunals.90 Nevertheless, he did not turn this majority of thought into a majority decision because, in his opinion, changes to the legal rules had to be made by a majority of opinion at the 'ultimate tribunal' rather than simply by a differently constituted bench.91 He concluded that: A decision to reverse turns upon more than theoretical correctness; it turns upon larger principles concerning the maintenance of the jurisprudence, judicial security and predictability. Included in those principles is, I believe, a practice for a judge to observe restraint in upholding his own dissent.92 Thus, because of the importance of predictability in icl, changes to the institutionalised legal rules become very difficult to accomplish, even where the 'uncritical reliance' on pre-existing jurisprudence raises concern for deficient performance of international criminal law justice.93

4.2
Potential for Reform These observations do not suggest that reform never happens in international legal institutions. But unless the institutional context changes drastically-as happened when the icc began operation and dismissed the jce theory used by the UN tribunals-any reform of the legal rules must be incremental, instead of revolutionary.
For example, even though the ac dismissed the Stakić tc's attempt to overthrow jce, the icty reformed jce to account for the different roles of, on the one hand, leadership figures and, on the other, persons on the ground. In 2007 the Brđanin ac found that the direct physical perpetrator of the crime did not have to be a member of the jce in order to impute criminal responsibility for that crime to all jce members.94 Instead, it was sufficient that the crime formed part of the common purpose and that one of the jce members was linked to the physical perpetrator and used the latter as a tool to commit the crime.95 Van Sliedregt points out that the Brđanin ac's version 20 of jce somewhat resembled the Stakić notion of co-perpetratorship,96 but it remained premised on the idea of a joint criminal enterprise. Consequently, even though the Stakić tc's attempt to radically change the rules, which the icty relied upon to assess principal liability, failed, the tribunal nevertheless transformed those rules over time.
It is possible that we may see a similar development at the icc with respect to the 'control-over-the-crime' theory. There have been recent indications that some chambers at the Court have sought to deal away with the abstract requirements of the 'control-over-the-crime' theory. Specifically, in the 2019 Yekatom and Ngaïssona decision on the confirmation of charges the judges did not examine whether the evidence established the existence of a 'common plan' between the suspects, as the 'control-over-the-crime' theory would have required, but simply assessed the available evidence regarding the link between the accused and each of the alleged criminal incidents.97 The Yekatom and Ngaïssona decision indicates the willingness of some icc judges to: first, simplify the rules for assessing criminal responsibility by dismissing any abstract theoretical concepts and, second, to focus on what the facts and evidence reveal about the crimes rather than on making those facts fit a pre-established theoretical framework. In fact, The Yekatom and Ngaïssona ptc judges referred to the concerns expressed earlier by Judge Van den Wyngaert with respect to the 'common plan' requirement of the 'control-over-the-crime' theory, when explaining their decision to focus specifically on the link between the accused and the criminal incidents, rather than the common plan.98 More recently, the Abd-Al-Rahman ptc took a similar approach. The judges noted that despite of the ubiquity of the concept of the 'common plan' pursuant to the 'control-over-the-crime' theory in icc jurisprudence, the compatibility of that concept with the statutory framework and its 'usefulness' was 'far from being a foregone conclusion' .99 The judges abstained from going into further deliberations on that topic, citing the limited functions of the pre-trial chamber, namely to determine whether the accused should be sent to trial. But they, nevertheless, displayed certain restraint towards the excessive reliance on the rigid requirements of the 'control' theory, such as the 'common plan' and 'essential contribution' elements. Specifically, the ptc noted that to require the judges to follow a particular terminology, including that of the 'control' theory, 'would be tantamount to favour form over substance' .100 Consequently, 96 Van Sliedregt, supra note 43, p. 170. 97 Yekatom and Ngaïssona Confirmation Decision, supra note 10, para. 60. 98 Ibid., footnote 106. 99 Abd-Al-Rahman Confirmation Decision, supra note 11, para. 44. 100 Ibid., para. 45. focusing on the substance of the case, the Abd-Al-Rahman ptc proceeded with an analysis of the facts that linked the accused with the criminal incidents, without making any explicit references to the existence of a 'common plan' between him and his allies, or to his 'essential contribution' to the crimes. This recent tendency of icc chambers to refrain from relying on the 'control-over-the-crime' terminology to frame the facts of the case has already presented a challenge with respect to the predictability of legal principles. After the Yekatom and Ngaïssona ptc confirmed the charges against Alfred Yekatom under Article 25(3)(a) using the new approach, the accused appealed the decision, arguing that he had not been provided with sufficient notice about the scope of the charges against him. Specifically, the accused submitted that the ptc had not expressly linked the facts of the case to the elements of co-perpetration, including the 'common plan' and the accused's 'essential contribution' to that plan.101 Thus, the Defence considered the new approach to assessing charges under Article 25(3)(a) too broad to provide fair notice about the charges.
Yet, the ac rejected the appeal. The judges found that the Regulations of the Court required that the legal characterisations of the facts must directly relate only to the statutory text, and not 'to other sources for further elaboration' .102 In the case of Article 25(3)(a) this means specifying the form of participation in a crime that the accused is charged with, as stated in the Rome Statute (namely, committing a crime 'individually, jointly with another person, or through another person').103 According to the ac, the pre-trial chambers were under no formal obligation to use any particular terminology aside from the statutory language.104 This line of reasoning of the ac is convincing: the Rome Statute lists the applicable modes of liability in significant detail, which should provide fair notice about the charges without further elaboration through the use of criminal law theory.
What is more intriguing is the ac's implicit downplaying of the role that the 'control-over-the-crime' theory has played in icc jurisprudence. The ac noted that the 'jointly with another' form of commission under Article 25(3)(a) 'can take a number of forms' depending on the case.105 The judges further noted that the icc chambers have relied on a 'broad spectrum' of approaches to confirm the charges and that, while some decisions have used 101 Yekatom and Ngaïssona Appeals Decision, supra note 12, para. 57. 102 Ibid., para. 37. 103 Ibid., para. 43. 104 Ibid., para. 60. 105 Ibid., para. 58. .107 Yet, unlike the Yekatom and Ngaïssona confirmation decision, the chambers in both Katanga and Ngudjolo and Gbagbo had earlier in the aforementioned decisions explicitly invoked the 'control-over-the-crime' terminology to analyse the facts of the cases,108 including the disputed notion of the 'common plan' .109 In fact, the Katanga and Ngudjolo ptc was the chamber that created the doctrine of 'indirect co-perpetration' based on the 'control-over-the-crime' theory adopted earlier in Lubanga. The Court's reliance on the 'control-overthe-crime' theory in cases involving Article 25(3)(a) charges has been so extensive, that the Yekatom and Ngaïssona ptc's decision to analyse the facts without explicit reference to that theory deeply surprised some commentators.110

4.3
Continuing the Reform? While the 'control-over-the-crime' theory has always been disputed at the icc, the voices of concern have formed the minority position, generally expressed in separate opinions, such as those of Judge Van den Wyngaert, Judge Fulford, Judge Morrison, and Judge Eboe-Osuji. Because the open criticism of the 'control' theory has always appeared in separate opinions, the role of that theory in icc jurisprudence might seem generally stable. However, as noted, the transformation of legal norms is an incremental process. Change is already taking place in icc jurisprudence, and it is most observable in those decisions that do not directly reject the 'control-over-the-crime' theory, but quietly refrain from using it.
It is still far from clear whether the newfound emphasis on the statutory language of Article 25(3)(a) or the 'control-over-the-crime' theory will prevail in icc jurisprudence. Notably, both the Yekatom and Ngaïssona and the Abd-Al-Rahman ptc s have cited the limited powers of the pre-trial chamber as an excuse to abstain from discussing the desirability of following the 'controlover-the-crime' theory. Both ptc s have left to the trial chambers the final decision whether to rely on the 'control' theory for the judgments or to base their analyses on the plain language of the Statute.
Another possibility is that the prominence of Article 25 ( There is certain merit to the proposition that greater reliance on Article 25(3)(d) Rome Statute could help avoid some of the inefficiencies associated with the 'control-over-the-crime' theory. Such flexible forms of criminal liability seem more apt to account for the variation of modes of organisation behind mass atrocities, compared to the rigid requirements of the 'control' Even though the timing of that decision and the re-characterisation of the charges triggered much criticism,117 Article 25(3)(d) arguably presented a better description of Katanga's conduct compared to the 'control-over-the-crime' theory and Article 25(3)(a). It is possible that much of the controversy surrounding the judgment could have been avoided if Article 25(3)(a) had been dismissed earlier in the proceedings. As noted by one expert commentator, the takeaway from the Katanga case was that 'squeezing ambitious fact patterns into legal categories' might 'backfire' in practice. 118 As discussed in section 2, however, reforms in icl require a certain degree of consensus among the icc judges, and preferably-support by the broader international legal community. Making Article 25(3)(a) less influential in icc jurisprudence and increasing the prominence of Article 25(3)(d) requires a relative consensus on two points: first, that provisions 25 (3) A certain level of consensus on the ranking of modes of liability according to degrees of blameworthiness appeared to be accomplished after several years of debates within and outside the Court. Even though several icc chambers have proposed that the difference between principal liability under Article 25(3)(a) and accessory liability under Article 25(3)(b)-(d) was 'not merely terminological' but carried normative significance,119 that argument has been often criticised by both icc judges120 and academics.121 The latter share the view that Article 25(3) Rome Statute merely lists a variety of forms of participation in an international crime, without attaching normative significance to one form or another.122 Notably, in 2014 the Katanga tc rejected the suggestion that Article 25(3) constituted a hierarchy of blameworthiness,123 thus, reforming the rule previously followed by the Lubanga tc. As observed by Van Sliedregt, the judges who have challenged the normative differentiation between modes of liability in the Rome Statute, including Judge Morrison and Judge Eboe-Osuji in Ntaganda, usually draw on their previous experience in domestic jurisdictions, that do not classify accomplice liability as 'less blameworthy' than principal liability.124 The flexibility of the latter approach appears particularly appealing, considering the difficulties that the 'control-over-the-crime' theory has experienced in differentiating between the roles and responsibilities of accused at different levels of the criminal apparatus. For example, in 2021 Dominic Ongwen, a brigade commander at the lra, was convicted at the icc under Article 25(3)(a) for committing some of the crimes listed in the charges through troops under his command.125 But the image of Ongwen as a principal perpetrator who had controlled part of the lra troops was hard to reconcile with that of the organization's infamous leader Joseph Kony who had controlled the decision-making  (3) is still occasionally challenged. In 2016 the Al-Mahdi tc once again suggested that as a 'general rule' the Rome Statute differentiated between principals and accessories, 'with principals bearing more blameworthiness' . 128 The fact that 7 years after the Katanga tc judgment, Judge Morrison and Judge Eboe-Osuji have made significant efforts to justify the view that Article 25(3) does not constitute a hierarchy of blameworthiness129 suggests that some of their icc colleagues still subscribe to the normative approach. The question has not been conclusively resolved and those who view the differentiation between modes of liability in the Rome Statute in a purely descriptive sense are still fighting to demonstrate the benefits of their approach and the inefficiency of the normative differentiation rule.
But even if most judges and legal experts agree that the rest of the modes of liability listed in Article 25(3) can be just as blameworthy as subparagraph (a), it is unclear whether Article 25(3)(d) in particular would gain a dominant role in icc jurisprudence. That provision has divided the opinions of legal experts ever since the drafting of the Rome Statue.130 Authoritative commentators have raised concerns about the provision's broad scope, and especially about its second alternative that criminalises contributions made merely 'in the 'control-over-the-crime' theory has shown that the increasing categorisation and systematisation of modes of liability might in theory avoid overexpansion of the culpability principle, but in practice could end up jeopardising legal reasoning.
Notably, the adoption of the 'control-over-the-crime' theory teaches a general lesson. The tendency to focus on criminal law theory and statutory provisions when debating how to improve legal reasoning in icl obfuscates the main challenge to delivering sound judgments, which is the difficulty of collecting evidence of the accused's involvement in the crimes when those persons had been removed from the scene of the crime. The most convincing judgments remain those, in which the accused's participation in the crimes is clearly demonstrated by the evidence. This usually happens in cases that concern a limited number of perpetrators and/or evidence of the accused's participation in hostilities. In those cases, it appears plausible to claim that the person had 'controlled' the commission of the crime or contributed to its commission by participating in a criminal enterprise. Conversely, the farther the accused from the scene of the crimes, the less convincing the claims of any legal theory become. The problems of the 'control-over-the-crime' theory in that regard were highlighted in the opinions of Judge Morrison and Judge Eboe-Osuji. The problems of jce had become subject to extensive scholarly analysis even earlier. As noted, Article 25(3)(d)-the prospective new dominant mode of liability-has also failed to avoid criticism.
Consequently, the deficiencies of the 'control-over-the-crime' theory evidenced in icc jurisprudence should not be interpreted merely as an indication of the shortcomings of a particular theory, but also as a cautionary tale about the danger of over relying on criminal law theory when it comes to international crimes. The lesson from icc jurisprudence should not be to search for a new criminal law theory in attempt to deliver more convincing judgments, but to appreciate the peculiarities of each case and to focus on the quality of the available evidence. The Yekatom and Ngaïssona and Abd-Al-Rahman cases make an important step in that regard.

Conclusion
The adoption and routinisation of rules of conduct should provide the benefits of clarity and predictability to the legal process. But as the case of the 'controlover-the-crime' theory demonstrates, a routinised rule is not necessarily an efficient rule. The 'control-over-the-crime' theory was meant to address some of the problems, which its predecessor jce faced, namely, the broad scope of minkova that doctrine and the lack of differentiation between the forms of participation in a jce. However, in practice the 'control-over-the-crime' theory has failed to address those problems, blurred the differences between the Court's cases, and led to abstract findings that are hard to reconcile with reality. But even though the 'control-over-the-crime' theory remains the dominant approach for establishing principal liability at the Court, there are indications that the criticism of this theory has been gradually paving the way towards reform. It might be too early to tell what the outcome of that reform will be, but there are a few possibilities. One option is dismissing the 'control-overthe-crime' theory as the default rule for interpreting Article 25(3)(a), as the Yekatom and Ngaïssona ac has done, and using it only in those situations where the facts of the case truly resemble the specific requirements of that theory. Another option is the one suggested by Judge Morrison-using Article 25(3)(a) only in cases of 'simple' criminality and relying on Article 25(3)(b), (c), and especially (d), in the bulk of icc cases.
Whichever way the icc chooses in the future, it is crucial to pay attention to the lessons learned from its experience with the 'control-over-the-crime' theory and refrain from prioritising the means (routinised rules concerning the modes of liability) over the end (sound legal judgment). Future will tell whether this disillusionment with the 'control-over-the-crime' theory will transform into disillusionment with criminal law theory in icl more generally.